Intel® Integrated Performance Primitives
Deliberate problems developing high-performance vision, signal, security, and storage applications.
6709 Discussions

DTMF detect issue

bigbaby
Beginner
‎03-16-2012 01:44 AM
744 Views

Hi,
Everybody! I want to use dtmf detect function of ipp, and I know this is provided through ipp-samples. But I found it has some problem.

First I run ipp-samples/speech-codec/build_intel64.sh to build
second, enter the directory: ipp-samples/speech-codecs/_bin/intel64_gcc4/bin, and here you will find "usc_tones".
Then, I get a pcm file(8000Hz, 16bit) a.pcm, which contain some DTMF digit signal
At last use this ./usc_tones -type DTMF -dt -f8000 a.pcm out", and output is "88505"

But the first "8" should not be detected, because this signal is too short( not more than 30ms )

So I check the file : ipp-samples/speech-codecs/codec/speech/td/src/usc_dtmf.c and add some "printf" in it

and do again, output is :

Start processing.
Input file: a.pcm
Output file: out

Tone detections.
### td/src/usc_dtmf.c: 313 pIppTDParams->dtmf_fs: 77
### td/src/usc_dtmf.c: 313 pIppTDParams->dtmf_fs: 154
### td/src/usc_dtmf.c: 313 pIppTDParams->dtmf_fs: 231
### td/src/usc_dtmf.c: 388 pIppTDParams->hlen(dtmf_fs): 231
### td/src/usc_dtmf.c: 395 pIppTDParams->hlen: 308
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 385
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 462
### td/src/usc_dtmf.c: 400 detect: 8 pIppTDParams->hlen: 462
### td/src/usc_dtmf.c: 411 pIppTDParams->hlen: 0
8### td/src/usc_dtmf.c: 395 pIppTDParams->hlen: 77
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 154
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 231
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 308
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 385
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 462
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 539
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 616
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 693
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 770
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 847
### td/src/usc_dtmf.c: 400 detect: 8 pIppTDParams->hlen: 847
### td/src/usc_dtmf.c: 411 pIppTDParams->hlen: 0
8### td/src/usc_dtmf.c: 395 pIppTDParams->hlen: 77
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 154
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 231
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 308
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 385
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 462
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 539
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 616
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 693
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 770
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 847
### td/src/usc_dtmf.c: 400 detect: 5 pIppTDParams->hlen: 847
### td/src/usc_dtmf.c: 411 pIppTDParams->hlen: 0
5### td/src/usc_dtmf.c: 395 pIppTDParams->hlen: 77
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 154
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 231
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 308
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 385
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 462
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 539
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 616
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 693
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 770
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 847
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 924
### td/src/usc_dtmf.c: 400 detect: 0 pIppTDParams->hlen: 924
### td/src/usc_dtmf.c: 411 pIppTDParams->hlen: 0
0### td/src/usc_dtmf.c: 395 pIppTDParams->hlen: 77
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 154
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 231
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 308
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 385
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 462
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 539
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 616
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 693
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 770
### td/src/usc_dtmf.c: 400 detect: 5 pIppTDParams->hlen: 770
### td/src/usc_dtmf.c: 411 pIppTDParams->hlen: 0
5
Finish.


We can see through output below, "hlen" is bigger than it should be: 462, and this cause the first digit '8' is detected.
I don't know why this var is initialized by dtmf_fs at biginning(file usc_dtmf.c line 387), but I beleive some problems exists here.

Are there some Intel guys can help me? Thanks !

### td/src/usc_dtmf.c: 313 pIppTDParams->dtmf_fs: 77
### td/src/usc_dtmf.c: 313 pIppTDParams->dtmf_fs: 154
### td/src/usc_dtmf.c: 313 pIppTDParams->dtmf_fs: 231
### td/src/usc_dtmf.c: 388 pIppTDParams->hlen(dtmf_fs): 231
### td/src/usc_dtmf.c: 395 pIppTDParams->hlen: 308
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 385
### td/src/usc_dtmf.c: 418 pIppTDParams->hlen: 462
### td/src/usc_dtmf.c: 400 detect: 8 pIppTDParams->hlen: 462

a.pcm
18 KB
0 Kudos

Link Copied

8 Replies
bigbaby
Beginner
‎03-19-2012 06:52 PM
744 Views
Nobody can help me ?
0 Kudos
Copy link
bigbaby
Beginner
‎03-20-2012 06:05 PM
744 Views
Nobody ?myself DING!
0 Kudos
Copy link
Chao_Y_Intel
Moderator
‎03-20-2012 06:36 PM
744 Views

Hello,

Thanks for the patience. I will check some information with some expert, and may provide some update.

Regards,
Chao

0 Kudos
Copy link
Chao_Y_Intel
Moderator
‎03-21-2012 12:30 AM
744 Views
Hi,

Here is some inputs for the problems:

The DTMF sample is used to demonstrate some usage of related IPP functions. It is developed according to the ITU Q.23 specification. It is very general and there arent any restrictions regarding tone length in that specification. Some restricted false detections are possible. If you have some specific usage, or improvement on that detection, it needs your further development with the source code.

Thanks,
Chao
0 Kudos
Copy link
bigbaby
Beginner
‎03-21-2012 05:40 AM
744 Views
HI,
Thanks for your replay, but I still feel disappoint with this.
To the sample code, I think it want to detect the tone which length is longer than 40ms.(hlen >= 320).
You really don't care about the bug in sample code ?
0 Kudos
Copy link
OKohl
Beginner
‎03-21-2012 06:29 AM
744 Views
I just have a few remarks (I'm not an Intel Employee):

1. Note that 'pIppTDParams->hlen' is given in bytes (not msec)
2. From a first glimpse of the code I would guess that 'pIppTDParams->hlen' is not set to 0
if 'toneCandidate' changes between two consecutive frames (not sure though).

good luck


0 Kudos
Copy link
bigbaby
Beginner
‎03-21-2012 06:46 PM
744 Views
thanks for your reply.

1 hlen is given in samples(2bytes), not bytes. So 320 means 40ms while sampling rate is 8000
2 I feel problem occurs at the beginning, hlen is set to a too big value, so the first digit will be mis-detected. But at the subsequence detection, it works well.
0 Kudos
Copy link
OKohl
Beginner
‎03-21-2012 11:52 PM
744 Views
Hey,

I miss understood something: At first you described the digit "8" was false-detected but it should not have been detected because it is too short (only 30 msec). Now you wrote the digit was miss-detected...which is it?

Could it be that the dtmf is simply not synchronized with the sampled frames (of ~30 msec),
so the first frame partialy contains dtmf 8 and the tone is not positivily detected, therefore only at the following frame it is detected?
0 Kudos
Copy link
Reply

Community support is provided during standard business hours (Monday to Friday 7AM - 5PM PST). Other contact methods are available here.

Intel does not verify all solutions, including but not limited to any file transfers that may appear in this community. Accordingly, Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

For more complete information about compiler optimizations, see our Optimization Notice.